The Agent Skills Directory

[SAFE]: The skill is a utility for managing local Markdown files. It provides clear, task-specific instructions for metadata maintenance without any suspicious or malicious operations.
[COMMAND_EXECUTION]: The skill utilizes allowed platform tools (Grep, Read, Edit, Write) to perform string replacements within files. There is no evidence of arbitrary shell command execution or the use of unsanitized user input in a command context.
[DATA_EXFILTRATION]: All operations are local to the file system. The skill does not initiate network requests or attempt to move data to external locations.
[PROMPT_INJECTION]: The instructions do not attempt to override agent safety protocols, extract system prompts, or bypass constraints. The language is professional and focused on the stated utility.
[CREDENTIALS_UNSAFE]: No hardcoded secrets, API keys, or sensitive credentials were found. The skill operates on notes and specifically avoids sensitive directories such as .git/ or .ssh/.
[INDIRECT_PROMPT_INJECTION]:
Ingestion points: The skill reads user-supplied Markdown files using Grep and Read tools.
Boundary markers: The instructions focus on structured YAML frontmatter sections, which naturally provides some isolation from the rest of the note content.
Capability inventory: The skill has the ability to write to and edit local files using Write and Edit tools.
Sanitization: The skill uses specific regex-like patterns (e.g., ^id:) to identify content for modification, reducing the risk of misinterpreting malicious instructions embedded in the notes.

vault-frontmatter